Multi-antenna techniques can significantly increase the data rates and reliability of a wireless communication system. The performance is particularly improved if both the transmitter and the receiver are equipped with multiple antennas, which results in a multiple-input multiple-output (MIMO) communication channel. Such systems and/or related techniques are commonly referred to as MIMO.
The Long Term Evolution (LTE) standard is currently evolving with enhanced MIMO support. A component in LTE is the support of MIMO antenna deployments and MIMO related techniques. Currently LTE-Advanced supports an 8-layer spatial multiplexing mode for 8 transmit (Tx) antennas with channel dependent precoding. The spatial multiplexing mode is aimed for high data rates in favorable channel conditions. An illustration of the spatial multiplexing operation 100 is provided in FIG. 1, where there are NT antenna 110 ports and NT inverse fast Fourier transformers (IFFTs) 120.
As seen, the information carrying symbol vector s 130 is multiplied by an NT×r precoder matrix W 140, which serves to distribute the transmit energy in a subspace of the NT (corresponding to NT antenna ports) dimensional vector space. The precoder matrix W 140 is typically selected from a codebook of possible precoder matrices, and typically indicated by means of a precoder matrix indicator (PMI), which specifies a unique precoder matrix in the codebook for a given number of symbol streams. The r symbols in s 130 each correspond to a layer 150 and r is referred to as the transmission rank. In this way, spatial multiplexing is achieved since multiple symbols can be transmitted simultaneously over the same time/frequency resource element (TFRE). The number of symbols r is typically adapted to suit the current channel properties.
LTE uses Orthogonal Frequency Division Multiplexing (OFDM) in the downlink (and Discrete Fourier Transform (DFT) precoded OFDM in the uplink) and hence the received NR×1 vector yn for a certain TFRE on subcarrier n (or alternatively data TFRE number n) is thus modeled byyn=HnWsn+en  Equation 1where en is a noise/interference vector obtained as realizations of a random process, and NR is the number of receive antennas. The precoder W can be a wideband precoder, which is constant over frequency, or frequency selective.
The precoder matrix W is often chosen to match the characteristics of the NR×NT MIMO channel matrix Hn, resulting in so-called channel dependent precoding. This is also commonly referred to as closed-loop precoding and essentially strives for focusing the transmit energy into a subspace which is strong in the sense of conveying much of the transmitted energy to the wireless device. In addition, the precoder matrix may also be selected to strive for orthogonalizing the channel, meaning that after proper linear equalization at the wireless device, the inter-layer interference is reduced.
One example method for a wireless device to select a precoder matrix W can be to select the Wk that maximizes the Frobenius norm of the hypothesized equivalent channel:
                              max          k                ⁢                                                                                          H                  ^                                n                            ⁢                              W                k                                                          F          2                                    Equation        ⁢                                  ⁢        2            Where Ĥn is a channel estimate, possibly derived from Channel State Information-Reference Signal (CSI-RS) as described below;Wk is a hypothesized precoder matrix with index k; andĤnWk is the hypothesized equivalent channel.
In closed-loop precoding for the LTE downlink, the wireless device transmits, based on channel measurements in the forward link (downlink), recommendations to the base station, e.g., eNodeB (eNB), of a suitable precoder to use. The base station configures the wireless device to provide feedback according to the wireless device's transmission mode, and may transmit CSI-RS and configure the wireless device to use measurements of CSI-RS to feedback recommended precoding matrices that the wireless device selects from a codebook. A single precoder that is supposed to cover a large bandwidth (wideband precoding) may be fed back. It may also be beneficial to match the frequency variations of the channel and instead feedback a frequency-selective precoding report, e.g., several precoders, one per subband. This is an example of the more general case of channel state information (CSI) feedback, which also encompasses feeding back other information than recommended precoders to assist the base station in subsequent transmissions to the wireless device. Such other information may include channel quality indicators (CQIs) as well as transmission rank indicator (RI).
Given the CSI feedback from the wireless device, the base station determines the transmission parameters it wishes to use to transmit to the wireless device, including the precoding matrix, transmission rank, and modulation and coding scheme (MCS). These transmission parameters may differ from the recommendations the wireless device makes. Therefore, a rank indicator and MCS may be signaled in downlink control information (DCI), and the precoding matrix can be signaled in DCI or the base station can transmit a demodulation reference signal from which the equivalent channel can be measured. The transmission rank, and thus the number of spatially multiplexed layers, is reflected in the number of columns of the precoder W. For efficient performance, it is important that a transmission rank that matches the channel properties is selected.
In LTE Release-10 (Rel-10), a new reference symbol sequence was introduced for the intent to estimate downlink channel state information, the CSI-RS. The CSI-RS provides several advantages over basing the CSI feedback on the common reference symbols (CRS) which were used, for that purpose, in previous releases. First, the CSI-RS is not used for demodulation of the data signal, and thus does not require the same density (i.e., the overhead of the CSI-RS is substantially less). Secondly, CSI-RS provides a much more flexible means to configure CSI feedback measurements (e.g., which CSI-RS resource to measure on can be configured in a wireless device specific manner).
By measuring a CSI-RS transmitted from the base station, a wireless device can estimate the effective channel the CSI-RS is traversing including the radio propagation channel and antenna gains. In more mathematical rigor, this implies that if a known CSI-RS signal x is transmitted, a wireless device can estimate the coupling between the transmitted signal and the received signal (i.e., the effective channel). Hence, if no virtualization is performed in the transmission, the received signal y can be expressed asY=Hx+e  Equation 3and the wireless device can estimate the effective channel H.
Up to eight CSI-RS ports can be configured in LTE Rel-10, that is, the wireless device can estimate the channel from up to eight transmit antennas.
Related to CSI-RS is the concept of zero-power CSI-RS resources (also known as a muted CSI-RS) that are configured just as regular CSI-RS resources, so that a wireless device knows that the data transmission is mapped around those resources. The intent of the zero-power CSI-RS resources is to enable the network to mute the transmission on the corresponding resources, in order to boost the Signal to Interference plus Noise Ratio (SINR) of a corresponding non-zero power CSI-RS, possibly transmitted in a neighbor cell/transmission point. For Release 11 (Rel-11) of LTE, a special zero-power CSI-RS was introduced that a wireless device is mandated to use for measuring interference plus noise. A wireless device can assume that the transmission points (TPs) of interest are not transmitting on the zero-power CSI-RS resource, and the received power can therefore be used as a measure of the interference plus noise.
Based on a specified CSI-RS resource and on an interference measurement configuration (e.g., a zero-power CSI-RS resource), the wireless device can estimate the effective channel and noise plus interference, and consequently also determine the rank, precoding matrix, and MCS to recommend to best match the particular channel.
Some installations are equipped with two dimensional antenna arrays and some of the presented embodiments use such antennas. Such antenna arrays may be (partly) described by the number of antenna columns corresponding to the horizontal dimension Nh, the number of antenna rows corresponding to the vertical dimension Nv, and the number of dimensions corresponding to different polarizations Np. The total number of antennas is thus N=NhNvNp. It should be pointed out that the concept of an antenna is non-limiting in the sense that it can refer to any virtualization (e.g., linear mapping) of the physical antenna elements. For example, pairs of physical sub-elements could be fed the same signal, and hence share the same virtualized antenna port.
An example of a 4×4 array with cross-polarized antenna elements 200 is shown in FIG. 2, where the horizontal dimension “1” represents Nh and the vertical dimension “m” represents the Nv.
Precoding may be interpreted as multiplying the signal with different beamforming weights for each antenna prior to transmission. A typical approach is to tailor the precoder to the antenna form factor, i.e., taking into account Nh, Nv and Np when designing the precoder codebook.
A common type of precoding is to use a DFT-precoder, where the precoder vector used to precode a single-layer transmission using a single-polarized uniform linear array (ULA) with N antennas is defined as
                    w                  1          ⁢          D                    ⁡              (        k        )              =                  1                  N                    ⁡              [                                                            e                                  j2                  ⁢                                                                          ⁢                                      π                    ·                    0                    ·                                          k                      QN                                                                                                                                              e                                  j                  ⁢                                                                          ⁢                  2                  ⁢                                      π                    ·                    1                    ·                                          k                      QN                                                                                                                              ⋮                                                                          e                                  j                  ⁢                                                                          ⁢                  2                  ⁢                                      π                    ·                                          (                                              N                        -                        1                                            )                                        ·                                          k                      QN                                                                                                          ]              ,where k=0,1, . . . QN−1 is the precoder index and Q is an integer oversampling factor. A corresponding precoder vector for a two-dimensional uniform planar array (UPA) can be created by taking the Kronecker product of two precoder vectors as w2D(k,l)=w1D(k)⊗w1D(l). Extending the precoder for a dual-polarized UPA may then be done as
            w                        2          ⁢          D                ,        DP              ⁡          (              k        ,                                  ⁢        l        ,                                  ⁢        ϕ            )        =                    [                                            1                                                                          e                                  j                  ⁢                                                                          ⁢                  ϕ                                                                    ]            ⊗                          ⁢                        w                      2            ⁢            D                          ⁡                  (                      k            ,                                                  ⁢            l                    )                      =                                                             [                                                                                                                  w                                                  2                          ⁢                          D                                                                    ⁡                                              (                                                  k                          ,                          l                                                )                                                                                                                                                                                e                                                  j                          ⁢                                                                                                          ⁢                          ϕ                                                                    ⁢                                                                        w                                                      2                            ⁢                            D                                                                          ⁡                                                  (                                                      k                            ,                            l                                                    )                                                                                                                                ]                        ⁡                          [                                                                                                                  w                                                  2                          ⁢                          D                                                                    ⁡                                              (                                                  k                          ,                          l                                                )                                                                                                  0                                                                                        0                                                                                                      w                                                  2                          ⁢                          D                                                                    ⁡                                              (                                                  k                          ,                          l                                                )                                                                                                        ]                                ⁡                      [                                                            1                                                                                                  e                                          j                      ⁢                                                                                          ⁢                      ϕ                                                                                            ]                          ,            where ejϕ is a co-phasing factor that may for instance be selected from the QPSK alphabet
  ϕ  ∈            {              0        ,                  π          2                ,        π        ,                              3            ⁢            π                    2                    }        .  A precoder matrix W2D,DP for multi-layer transmission may be created by appending columns of DFT precoder vectors asW2D,DP=[w2D,DP(k1,l1,ϕ1) w2D,DP(k2,l2,ϕ2) . . . w2D,DP(kR,lR,ϕR)],where R is the number of transmission layers, i.e., the transmission rank. In a common special case for a rank-2 DFT precoder, k1=k2=k and l1=l2=l, meaning that
          ⁢            W                        2          ⁢          D                ,        DP              =                  [                                            w                                                2                  ⁢                  D                                ,                DP                                      ⁡                          (                              k                ,                l                ,                                  ϕ                  1                                            )                                ⁢                                          ⁢                                          ⁢                                    w                                                2                  ⁢                  D                                ,                DP                                      ⁡                          (                              k                ,                l                ,                                  ϕ                  2                                            )                                      ]            =                                    [                                                                                                      w                                              2                        ⁢                        D                                                              ⁡                                          (                                              k                        ,                        l                                            )                                                                                        0                                                                              0                                                                                            w                                              2                        ⁢                        D                                                              ⁡                                          (                                              k                        ,                        l                                            )                                                                                            ]                    ⁡                      [                                                            1                                                  1                                                                                                  e                                          j                      ⁢                                                                                          ⁢                                              ϕ                        1                                                                                                                                  e                                          j                      ⁢                                                                                          ⁢                                              ϕ                        2                                                                                                                  ]                          .            
With multi-user MIMO, two or more users in the same cell are co-scheduled on the same time-frequency resource. That is, two or more independent data streams are transmitted to different wireless devices at the same time, and the spatial domain is used to separate the respective streams. By transmitting several streams simultaneously, the capacity of the system can be increased. This however, comes at the cost of reducing the SINR per stream, as the power has to be shared between streams and the streams will cause interference to each-other.
When increasing the antenna array size, the increased beamforming gain will lead to higher SINR, however, as the user throughput depends only logarithmically on the SINR (for large SINRs), it is instead beneficial to trade the gains in SINR for a multiplexing gain, which increases linearly with the number of multiplexed users.
Accurate CSI is required in order to perform appropriate nullforming between co-scheduled users. In the current LTE Release 13 (Rel-13) standard, no special CSI mode for MU-MIMO exists and thus, MU-MIMO scheduling and precoder construction has to be based on the existing CSI reporting designed for single-user MIMO (that is, a PMI indicating a DFT-based precoder, a RI and a CQI). This may prove quite challenging for MU-MIMO, as the reported precoder only contains information about the strongest channel direction for a user and may thus not contain enough information to do proper nullforming, which may lead to a large amount of interference between co-scheduled users, reducing the benefit of MU-MIMO.
A multi-beam precoder may be defined as a linear combination of several DFT precoder vectors as
            w      MB        =                  ∑        i            ⁢                        c          i                ·                              w                                          2                ⁢                D                            ,              DP                                ⁡                      (                                          k                i                            ,                              l                i                            ,                              ϕ                i                                      )                                ,where {ci} may be general complex coefficients. Such a multi-beam precoder may more accurately describe the wireless device's channel and may thus bring an additional performance benefit compared to a DFT precoder, especially for MU-MIMO where rich channel knowledge is desirable in order to perform nullforming between co-scheduled wireless devices.
Existing solutions for MU-MIMO based on implicit CSI reports with DFT-based precoders have problems with accurately estimating and reducing the interference between co-scheduled users, leading to poor MU-MIMO performance.
Multi-beam precoder schemes may lead to better MU-MIMO performance, but at the cost of increased CSI feedback overhead and wireless device precoder search complexity. It is an open problem of how an efficient multi-beam codebook that results in good MU-MIMO performance but low feedback overhead should be constructed, as well as how the CSI feedback should be derived by the wireless device.